TCA9548A: TCA9548A: mux channel selection issue / TCA9548A: i2c multiplexing problem

Hey Manikandan,

I don't see an attachment here but I've seen this problem pop up about 3 times in the past week involving the FTDI dongle and our I2C switches.

Here is an example where the customer's master is not properly sending a stop condition after the write transaction:

Can you verify that in your o-scope shot. After sending the I2C address and write the byte, you issue a stop condition?

Thanks,

-Bobby

Hey Bobby,

Thanks for the reply quick reply and link for expected scope pattern. Let me go through the link and issue i2c commands as required. As this is lock down scenario, it is taking little long time. Meanwhile, I'm attaching the schematics and captures as a zip file. 

It would be great, if you can guide in specific direction

ftdi_i2cmux_pmic.zip

Thanks,

Manikandan

Hey Bobby,

I could check the channel selection write and see  'stop' bit from i2c master write. But the stop bit arises ~500uS after ack of slave channel sel write with ~5uS time interval between SCL rise and SDA rise. (Attached waveform is for reference)

I would like to know,

Is there any time restriction to emit a stop condition and will the SCx toggle on pmic ack on a valid stop condition

Is it okay to focus only on achieving correct 'stop' condition 

Thanks Bobby for the support,

Manikandan

In scope capture, Ch3 shows as SC0 but actually it is probed on SC1.

Same verified in SC0 as well by channel selection as 0x01

"Is there any time restriction to emit a stop condition and will the SCx toggle on pmic ack on a valid stop condition"

The state machine of our device does not have an internal clock/oscillator so it does not actually know time. It is looking for a specific edge, That is while data line is low, clock rises, then data shortly after.

"Is it okay to focus only on achieving correct 'stop' condition"

After a valid right, yes.

Please note, we need a valid write transaction followed by a stop condition. If you try to do a repeated start and then do a read and stop condition, the channel may not enable.

Below, I went into detail in another post where I tried to recreate the stop condition from an FTDI dongle and showed that by modifying when I release the clock signal, I'm able to properly enable another channel even after 200ms of delay.

-Bobby